The prevalence and predictors of bipolar and borderline personality disorders comorbidity: Systematic review and meta-analysis

Author’s Accepted Manuscript The prevalence and predictors of bipolar and borderline personality disorders comorbidity: Systematic review and meta-analysis M. Fornaro, L. Orsolini, S. Marini, D. De Berardis, G. Perna, A. Valchera, L. Ganança, M. Solmi, N. Veronese, B. Stubbs www.elsevier.com/locate/jad

PII: DOI: Reference:

S0165-0327(15)31291-X http://dx.doi.org/10.1016/j.jad.2016.01.040 JAD8012

To appear in: Journal of Affective Disorders Received date: 18 November 2015 Revised date: 4 January 2016 Accepted date: 24 January 2016 Cite this article as: M. Fornaro, L. Orsolini, S. Marini, D. De Berardis, G. Perna, A. Valchera, L. Ganança, M. Solmi, N. Veronese and B. Stubbs, The prevalence and predictors of bipolar and borderline personality disorders comorbidity: Systematic review and meta-analysis, Journal of Affective Disorders, http://dx.doi.org/10.1016/j.jad.2016.01.040 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

The prevalence and predictors of Bipolar and Borderline Personality disorders comorbidity: systematic review and meta-analysis

Fornaro M. 1; Orsolini, L. 2,3,4; Marini S. 5; De Berardis D. 6; Perna G. 7; Valchera A.4; Ganança L.1,8; Solmi M.9,10; Veronese N11; Stubbs B 12,13

1. New York State Psychiatric Institute, Columbia University, NY, USA 2. School of Life and Medical Sciences, University of Hertfordshire, Hatfield, Herts, UK 3. Department of Psychiatry and Neuropsychology, Maastricht University, Maastricht, Netherlands 4. Hermanas Hospitalarias – Villa San Giuseppe, Ascoli Piceno, Italy 5. Department of Neuroscience & Imaging, “G. D’Annunzio” University, Chieti, Italy 6. National Health Service, Department of Mental Health, Psychiatric Service of Diagnosis and Treatment, Hospital “G. Mazzini”, ASL 4, Teramo, Italy 7. Hermanas Hospitalarias - Villa San Benedetto Menni Hospital, Department of Clinical Neurosciences, FoRiPsi 8. Department of Psychiatry, School of Medicine, University of Lisbon, Portugal 9. Department of Neuroscience, University of Padova, Padua, Italy 10. National Health Care System, Padua Local Unit ULSS 17, Italy 11. Department of Medicine- DIMED, Geriatrics Section, University of Padua, Italy 12. Physiotherapy Department, South London and Maudsley NHS Foundation Trust, London SE5 8AZ, UK 13. Health Service and Population Research Department, Institute of Psychiatry, King's College London, De Crespigny Park, London SE5 8AF, UK

Michele FORNARO, MD, PhD: [email protected] Laura ORSOLINI, MD: [email protected] Stefano MARINI, MD: [email protected] Domenico DE BERARDIS, MD, PhD: [email protected] Giampaolo PERNA, MD, PhD [email protected] Alessandro VALCHERA, MD [email protected] Licinia GANANÇA, MD [email protected] Marco SOLMI, MD [email protected] Nicola VERONESE, MD [email protected] Brendon STUBBS, MSc, PhD: [email protected]

Corresponding author: Michele Fornaro, Columbia University, New York State Psychiatric Institute, 1051 Riverside Drive, Unit 42, room 2729, NYC, ZIP 10032. Tel. +1(646) 774-5000. Email: [email protected]

Abstract Introduction: Data about the prevalence of borderline personality (BPD) and bipolar (BD) disorders comorbidity are scarce and the boundaries remain controversial. We conducted a systematic review and meta-analysis investigating the prevalence of BPD in BD and BD in people with BPD.

Methods: Two independent authors searched MEDLINE, Embase, PsycINFO and the Cochrane Library from inception till November 4 2015. Articles reporting the prevalence of BPD and BD were included. A random effects meta-analysis and meta-regression were conducted. Results: Overall, 42 papers were included: 28 considering BPD in BD and 14 considering BD in BPD. The trim and fill adjusted analysis demonstrated the prevalence of BPD among 5,273 people with BD (39.94±11.78 years, 44% males) was 21.6% (95% CI 17.0-27.1). Higher comorbid BPD in BD were noted in BD II participants (37.7%, 95% CI 21.9-56.6, studies=6) and North American studies (26.2%, 95% CI 18.7-35.3, studies=11). Meta regression established that a higher percentage of males and higher mean age significantly (p<0.05) predicted a lower prevalence of comorbid BPD in BD participants. The trim and fill adjusted prevalence of BD among 1,814 people with BPD (32.22±7.35 years, 21.5% male) was 18.5% (95% CI 12.7-26.1). Limitations: Paucity of longitudinal/control group studies and accurate treatment records. Conclusions: BPD-BD comorbidity is common, with approximately one in five people experiencing a comorbid diagnosis. Based on current diagnostic constructs, and a critical interpretation of results, both qualitative and quantitative syntheses of the evidence prompt out the relevance of differences rather similarities between BD and BPD.

Keywords: Bipolar disorder (BD); borderline personality disorder (BPD); comorbidity; prevalence; predictors; systematic review; meta-analysis.

1. Introduction Discriminating between borderline personality (BPD) and bipolar (BD) disorder is difficult (Barroilhet et al., 2013; Ghaemi et al., 2014), as well as crucial in the clinical and critical evaluation of the comorbidity rates between the twos (Zimmerman and Morgan, 2013). The odds of confusing BPD with BD are particularly high for severe bipolar cases (Ghaemi 2

and Barroilhet, 2015), essentially due to differential emphasis placed on similarities rather than differences between the twos (Agius et al., 2012; Ghaemi et al., 2014; Vieta and Suppes, 2008; Zimmerman and Morgan, 2013). A bio-psychosocial approach promoting an explanatory psychological effect of a biological (cyclothymic) temperament in the understanding of the controversies surrounding BD-BPD nosological dilemmas has not been unanimously accepted, tough being ontologically and clinically suggestive (Khalili, 2014). Moreover, BPD and BD share substantial overlap in the nosological validator of mood lability, especially for currently depressed BD Type-II (BD-II) cases (Henry et al., 2001), “soft bipolar” atypical forms of depressions sharing a common cyclothymic temperament diathesis (Perugi et al., 2011; Perugi et al., 2003) and “ultra-rapid” (Mackinnon and Pies, 2006), “stably instable” bipolar cases (Akiskal, 1994). Mood lability is a Diagnostic and Statistical Manual for Mental Disorders, Fifth Edition (DSM-5) (APA, 2013) criterion for BPD, but not BD, though being common also in this latter (Goodwin and Jamison, 2007). DSM-defined atypical or manic features are infrequent in BPD compared to BD, though high rates of mixed features have been documented by large-sampled cross-sectional studies on major depressive episode patients (either DSM-defined “unipolar” or “bipolar” cases) with both BD and BPD according to permissive definitions (Allen et al., 2012; Perugi et al., 2013; Perugi et al., 2015; Young et al., 2012). Similarly, the symptom of impulsivity is closely allied to mood lability, and it is often seen as manifesting as sexual impulsivity in both BPD and BD, although it can also be physical, aggressive, financial (Ghaemi et al., 2014) or binge eating-related (Nagata et al., 2013; Perugi and Akiskal, 2002b). The affective lability of BPD vs. BD-II/cyclothymic patients nonetheless shows differential frequency and intensity patterns using both self-report and clinicianadministered measures (Reich et al., 2012; Swann et al., 2013). BD and BPD differ notably on a number of diagnostic validators, especially the course of illness of past sexual abuse (Bayes et

3

al., 2015; Briere and Elliott, 2003; Conus et al., 2010; Fossati et al., 1999; Maniglio, 2014) and history of para-suicidal self-harm (Joyce et al., 2010; Nock and Kessler, 2006). Genetic validators, treatment response, and neurobiological differences are also consistent between the twos (Ghaemi et al., 2014). Unsurprisingly, stating the controversy surrounding the relationship between BD and BPD, the most studied question concerns their actual diagnostic concordance, not only for independent samples but also for comorbid BD-BPD cases (Zimmerman and Morgan, 2013). Across studies, approximately 10% of BPD patients had BD-I, an additional 10% had BD-II. Likewise, approximately 20% of BD-II patients were diagnosed with BPD (Zimmerman and Morgan, 2013). While the comorbidity rates are substantial, each disorder is nonetheless usually diagnosed in the absence of the other across the studies, whereas studies directly comparing patients with BPD to BD cases (or BPD to BD) found significant differences over a broad-range of validators, actually challenging the notion of BPD being part of the broad bipolar spectrum (Zimmerman and Morgan, 2013). On the other side, while the “pragmatic approach” of the DSM-IV (and DSM-5) aims at reducing the rates of over-diagnosis, the “strict” validity of some diagnostic categories as BD and BPD has been questioned (Stein et al., 2010), meaning that it cannot be granted that BD and BPD necessarily represent clear-cut distinct diagnostic entities. In this view, the absence of any DSM guidance soliciting the assessment of BPD in BD or the opposite could ultimately lead to underestimation of comorbidity rates between the two. To our knowledge, no meta-analysis has specifically investigated BPD and BD comorbidity and predictors. Adopting a meta-analytic approach would therefore provide the advantage of pooling data from numerous studies in a logical manner towards a more accurate effect size which is closer to the true prevalence than when individual studies are considered separately (Ioannidis, 2009). In contrast to previous meta-analytic reports assessing a broad range of mood disorders comorbid with varying personality disorders (Friborg et al., 2014), the present systematic review

4

and meta-analysis, first of its kind to best of our knowledge, rather focuses on the prevalence and predictors of comorbid BPD↔BD in adults.

5

2. Materials and methods The present meta-analysis was conducted according to the Meta-analysis Of Observational Studies in Epidemiology (MOOSE) guidelines (Stroup et al., 2000), and the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Liberati et al.).

2.1 Information sources and search strategy Two independent authors searched MEDLINE, Scopus, Embase, PsycINFO and the Cochrane Library. The search strategy combined free text terms and exploded MESH headings for the topics of bipolar disorder and borderline personality disorder as following: (((((((Bipolar disorder) OR BD) OR Bipolar) OR Manic depressive disorder) OR Manic depressive) OR Manic)) AND (((Borderline personality disorder) OR Borderline) OR BPD). This latter MEDLINE strategy was then adapted for use in the other databases (Appendix A). Studies published in English through November 4 2015 were included. We further assessed the reference listing of retrieved relevant articles for potential inclusion of additional contributes.

2.2 Inclusion criteria 2.2.1 Study population and study design We considered studies that included comorbid cases of BD and BPD providing accurate diagnostic definitions based on either the DSM or the International Classification of Diseases ICD (any edition or text revision). Accounted BD populations at study included either BD-I, BDII and/or BD-Not Otherwise Specified (BD-NOS) cases. Participants of both sexes, 18 years of age or older were considered. Both population-based and hospital-based studies were included. Among hospital-based studies, inpatients, day-hospital and outpatient subjects were included; emergency care records excluded as considered non-representative. All experimental and observational study designs were included apart from case reports, opinion articles/letters to the Editor or conference proceedings

6

or reviews.

2.2.2 Outcome measures Primary outcomes were i) lifetime prevalence of comorbid BPD in BD patients and ii) lifetime prevalence of comorbid BD in BPD patients.

2.2.3 Study selection and data extraction Identified studies were independently reviewed for eligibility by three authors (MF, LO, SM) in a two-step based process; a first screening was performed based on title and abstract while full texts were retrieved for the second screening. Disagreements by reviewers were resolved by consensus at both stages. Data were extracted by two authors (BS, MF) and supervised by six additional authors (AV, LG, GP, DDB, MS and NV) using a purpose built data extraction spreadsheet. The data extraction spreadsheet was piloted on 10 randomly selected papers and modified accordingly. Both auto- and hand-searches for “type-I” (“duplicates among/across different databases”) and “type-II” (“duplicate publications in different Journals/issues”) (Qi et al., 2013) were performed based on Thompson Endnote X7® for Microsoft Windows®. Specifically, the recorded variables included the followings: author(s), year publication, study design, sample size, eventual follow-up or control group, socio-demographic status, concurrent psychotherapy (any type) or history of drug and/or physical treatment, outcome measures, conclusions, limitations, quality score, and quality differentiation. Whenever documented and reliably defined and/or most likely ascertainable by included original sources, accounted moderators focused on course of illness and pivotal variables reported in the literature with regard to the BD and BPD controversial boundaries, namely mood lability, impulsivity, history of childhood sexual abuse and/or lifetime physical self-harm.

2.2.4 Quality assessment

7

Two authors (LO and SM) independently assessed the quality of selected studies using the checklist developed by Downs and Black both for randomized and non- randomized studies (Downs and Black, 1998), as reported in tables 1 and 2 for those studies included in our systematic review. Eventual disagreements by reviewers were resolved by consensus.

2.2.5 Meta-analysis We pooled individual study data using DerSimonian-Laird proportion method (DerSimonian and Laird, 1986) with Comprehensive Meta-Analysis® software (version 3). Due to anticipated heterogeneity, a random effects meta-analysis was employed. First, we calculated the prevalence of BPD comorbidity in BD participants. Second, we calculated the prevalence of BD participants in people with BPD. For both meta-analyses, when three or more studies were available, we conducted sub group meta-analyses investigating BPD-BD comorbidity according to geographical region, study setting (inpatient versus outpatient/ community), and BD classification (BD-I versus BD-II and mixed). We also anticipated on conducting separate pooled prevalence of BD according to gender and classification of BD itself. We assessed publication bias with the visual inspection of a funnel plot (Higgins and Green, 2011) and the Begg (Begg and Mazumdar, 1994) and Egger (Egger et al., 1997) tests. In addition, for the main prevalence analysis we conducted a trim and fill adjusted analysis to remove the most extreme small studies from the positive side of the funnel plot, re-computing the effect size at each iteration, until the funnel plot is symmetric about the (new) effect size (Duval and Tweedie, 2000). Finally, we conducted several meta-regression analyses (if N≥3) to investigate potential moderators (age, percentage males and study population, as well as the pivotal differential features expected to differentiate between BD and BPD) with Comprehensive Meta-Analysis® (version 3).

8

3. Results One thousand four hundred twenty-four potential contributes were identified from searching the selected databases and listing references of relevant articles. After removing duplicates, 453 articles were retrieved. Studies were screened and selected on the basis of pre-specified inclusion and exclusion criteria (figure 1). Forty-two articles were ultimately included in the systematic review: 28 articles about comorbid BPD in BD and 14 articles about comorbid BD in BPD. Out the included 42 studies, only 3 (Alnaes and Torgersen, 1988; Comtois et al., 1999; Zimmerman and Mattia, 1999) concurrently assessed the prevalence rates and clinical features associated to either of BD in BPD cases or the opposite (please refer to tables 1 and 2 respectively for additional details and referencing). Figure 1 here 3.1 Comorbid BPD in BD 3.1.1 Study characteristics The characteristics of included studies are reported in table 1. Eight studies were cross-sectional studies, 7 were case-control, 7 prospective cohort studies, 4 cohort studies and 2 clinical trials. Two (7.14%) studies were population-based while 26 (92.86%) were hospital-based ones. A total of 5,273 BD patients were represented among the 28 included studies. The mean age of BD patients was 39.94±11.78 years and 44% were males. As detailed in table 1, twelve studies were carried in Europe, 11 in North America, 3 in South America and 2 (Perugi et al., 2013; Perugi et al., 2015) across Europe, Asia and Africa.

3.1.2 Quality of results The average score of the studies assessing BPD comorbidity in BD (n=28) was 23.43 out of 31 (range=11-25). Please refer to table 1 for details about the given studies.

3.1.3 Meta-analysis of the pooled prevalence of BPD in BD patients

9

The pooled prevalence of BPD comorbidity among 5,273 people with BD over 28 studies was 18.6% (95% CI 14.44-23.71, I2=91%) (figure 2). Whilst the Egger test did not indicate any publication bias (-0.88, p=0.33) the trim and fil analysis adjusted five studies and calculated a new prevalence of BPD comorbidity at 21.6% (95% CI 17.0-27.1). Figure 2 here 3.2 Sub group meta-analyses All sub group analyses results investigating BPD and BD comorbidity are presented in table 3 but will briefly be explored.

3.2.1 BPD prevalence according to the BD type BPD prevalence among three studies conducted in BD-I patients was 12.5% (95% CI 6.97-21.27, n=422), and BPD prevalence was 27.3% in BD-II participants (95% CI 16.6-41.3%, n=377). The prevalence of BPD among studies conducted in multi-diagnostic subsets of BD patients (BD-I, BD-II, BD-NOS) was 18.5% (95% ci 13.1-25.3%, n=4474). The between group difference in prevalence was not significant (p=0.1). Whilst none of the pooled prevalence demonstrated any evidence of publication bias (see table 2 for Egger test and p values), the prevalence of BPD comorbidity increased in each BD type after the adjustment for publication. Notably, the prevalence of BPD comorbidity was 37.7% (95% CI 21.9-56.6) in BD-II participants.

3.2.2 Geographical variations in the prevalence of BPD The prevalence of BPD was 16.5% (95% CI 10.91 -24.29, studies 12, n=880) in Europe and 22.1% (95% CI 14.54 - 32.04, studies 11, n=2059) in North America (see table 3). The adjustment for publication bias in the trim and fill analysis resulted in the prevalence of BPD increasing to 26.2% (95% CI 18.7-35.3) in North America BD participant studies.

10

3.2.3 BPD prevalence according to the study setting No between group differences were observed in the prevalence of BPD in BD according to study setting (see table 3), although the highest prevalence was observed in outpatient settings (20.8%, 95% CI 15.2-27.8, studies 20, n=3067).

3.1.3 Predictors of the prevalence of BPD in BD patients Meta regression analysis demonstrated that a higher percentage of males predicted a lower prevalence of comorbid BPD in BD participants (β=-0.0581, 95% CI -0.0848 - -0.0315, p<0.001, studies=20) and explained over half of the between study heterogeneity (R² = 0.58). A higher mean age also predicted a lower prevalence of BPD comorbidity in BD participants (β=-0.0782, 95% CI -0.1569--0.0006, p=0.04, R² = 0.15, studies=20). The year of study publication was not related to the prevalence of BPD across the studies (β=-0.0007, 95% CI -0.0017-0.0003, p=0.19, R² = 0.07, studies=28).

3.3 Comorbid BD in BPD 3.3.1 Study characteristics The characteristics of included studies are reported in table 2. Eight studies were cross-sectional studies and 6 prospective cohort studies. All studies were hospital-based ones. A total of 1,814 BPD patients were represented among the 14 included studies. The mean age of BPD patients was 32.22±7.35 years and 21.5% were males. As detailed in table 2, one study was conducted in Europe, 12 in North America, and 1 (Gunderson et al., 2014) across Europe and North America.

3.3.2 Quality of results The average score of the studies assessing BD comorbidity in BPD (n=14) was 27.57 out of 31 (range=9-24). Please refer to table 2 for additional details.

11

3.3.3 Meta-analysis of the pooled prevalence of BD comorbidity in BPD patients The pooled prevalence of BD among 1,814 people with BPD was 16.58% (95% CI 11.38-23.53, I2=87%, Studies 14) (figure 3). Whilst no publication bias was evident (Egger=0.85, p=0.61), the trim and fill analysis adjusted for publication bias demonstrated the prevalence of BD in BPD was slightly higher at 18.5% (95% CI 12.7-26.1). Figure 3 here 3.4 Sub group meta-analyses Full details of all sub group analyses are presented in table 4 but will briefly be summarized.

4.4.4 Prevalence of BD according to different BD type BD mixed was evident in 19.89% (95% CI 12.23-30.67) of people with BPD, which was higher than BD I (15.30%, 95% CI 6.47-32.06) and BD II (12.65%, 95% CI 4.79-29.47), although the between group differences were not significantly different (p=0.24).

3.4.1 Geographical variations in the prevalence of BD The prevalence of BD across 12 studies in North America was 17.50% (95% CI 11.7025.36) which slightly increased after the adjustment for publication bias (18.8%, 95% CI 12.527.1). Only one study was conducted in Europe and another study recruited participants from North America and Europe (see table 4).

3.4.2 Prevalence of BD according to setting of the study Nine studies reported the prevalence of BD in BPD participants in outpatient settings with the pooled prevalence at 19.34% (95% CI 11.96-29.72). No significant differences were observed in other settings (see table 4).

3.3.3 Predictors of the prevalence of BD in BPD patients

12

Meta regression analysis demonstrated mean age moderated a higher prevalence of BD in BPD participants (β=0.292, 95% CI 0.2019 - 0.382, p<0.001, R² = 1.00) although only restricted to four studies. The year the study was published (β=0.0394, 95% CI -0.0178-0.0966, p=0.17, R² =0.05, studies=14) and the percentage of males (β= -0.0136, 95% CI -0.083-0.0558, p=0.77, R² =0.01, studies=5).

13

4. Discussion The current systematic review established that comorbid BPD among people with BD and BD among people with BPD is common. Specifically, in the first meta-analysis of its kind we found that after the adjustment for potential outliers the prevalence of BPD among 5,273 people with BD was 21.6% (95% CI 17.0-27.1). We established that 18.5% of people with BPD have a comorbid BD diagnosis. These prevalence rates are broadly concordant with a previous narrative systematic-review which found that approximately 10% of patients with BPD had BD-I and another 10 % had BD-II (roughly, a total of about 20% of patients with BPD were diagnosed with comorbid BD) (Zimmerman and Morgan, 2013). Likewise, this is also concordant with the previous qualitative synthesis documenting approximately 20% of BD-II patients diagnosed with comorbid BPD, though only 10 % of BD-I patients were diagnosed with comorbid BPD (Zimmerman and Morgan, 2013). Not only did we extend the literature by conducting the first formal meta-analysis, but also we were able to investigate potential moderators and subgroup differences in BD and BPD comorbidity. For instance, after the adjustment for publication bias we found higher rates of BPD in BD-II participants (37.7%, 95% CI 21.9-56.6) and North America studies (26.2%, 95% CI 18.7-35.3). Meta-regression analysis demonstrated that a higher percentage of males and increasing age predicted a lower comorbid prevalence of BPD in people with BD (i.e. more common in females and older adults). This latter trend is also in line with previous evidence pointing out higher rates of female patients, and younger mean age of onset, in those comorbid cases of either BD in BPD or BPD in BD vs. BD samples without comorbid BPD (Barbato and Hafner, 1998; McDermid et al., 2015; Sajatovic et al., 2006), despite the only meta-analytic study assessing the prevalence rates of personality disorders in mood disorder samples failed to provide sufficient information on the matter with respect to the BD↔BPD comorbidity to allow reliable comparisons (Friborg et al., 2014).

14

4.1 Critical evaluation of the synthesis and implications for the clinical practice and future lines of research “Comorbidity” is a term originally introduced to refer to the coexistence of two essentially independent and distinct disorders (Amerio et al., 2015; Fountoulakis, 2014; Maj, 2005). According to this early concept, there exists an “index” or “primary” disorder and a comorbid separate second disorder which potentially affects the selection of treatment and the prognosis of the index one (Feinstein, 1970). In Feinstein’s formulation, the implication was that a completely different and independent disease occurred at the same time as another disease. These two diseases co-occurred, more often than not, randomly (Amerio et al., 2015). On the contrary, the DSM explicitly produces overlapping clinical criteria for many diagnoses, allowing comorbidity in quite a different sense than in the medical meaning of the term as co-occurrence of independent diseases (Maj, 2005). Using DSM definition, it is unclear whether concomitant diagnoses actually reflect the presence of distinct clinical entities or refer to multiple manifestations of a single clinical entity (Maj, 2005). Indeed, this is a critical issue also with respect to the BD↔BPD comorbidity, which is still source of vivid debate and opposite views about the actual overlap (Perugi et al., 2011) or distinction (Ghaemi and Barroilhet, 2015; Ghaemi et al., 2014) existing between the twos, especially in case of Type-II bipolar depression (Bayes et al., 2015). Moreover, the DSM fail to provide any rule soliciting clinicians or researchers to explore the possibility of a comorbid personality disorder in BD, nor conversely to include or exclude BD as an explanation for the emotional instability (and often overt mood swings) seen in BPD patients (Henry et al., 2012). The distribution of the affective-emotional lability continuum features across BD (especially BD-II depression with atypical features and/or cyclothymic temperament) and BPD is not bimodal, as this is the case of anxious-sensitivity and impulsivity – though

15

presenting with peculiar neuropsychological differences in BPD vs. BD (Akiskal, 2004; Perugi and Akiskal, 2002a) and that even the most “cautious” authors devoted to the study of the boundaries/interface between BPD and “border” disorders (namely BD) acknowledge at least a partial overlap or continuum between the two conditions (Paris et al., 2007). This is a crucial issue to bear in mind in the interpretation of the quantitative synthesis provided by the present meta-analytic study, as this may actually be on the lower side, and it may also represent a relevant issue for the clinical practice as well hinder the search for reliable and valid endophenotype as well as other genotypic vulnerability markers for BD↔BPD interface (Siever et al., 2002), similarly to what would occur in case of BD underestimation in MDD cases (Fornaro et al., 2011). Among other consequences, an underestimation of BD comorbidity rates in BPD cases would lead to delayed comprehensive diagnosis and multi-disciplinary approach. This is compelling, considering that real world date indicate that a considerably high number of BPD patients receive psychotropic medications, often as part of polypharmacy regimens, including antidepressant monotherapy despite the absence of any therapeutic guideline endorsing such practice (Bridler et al., 2015). Therefore, underestimation of BPD-BD interface would lead to increased risk of exposure towards improper or harmful medications too. Furthermore, comparisons of BPD vs. BD samples (or the opposite) are almost invariably done by following a cross-sectional rather than a longitudinal approach recommended by Kraepelin in the assessment of manic-depressive illness (Angst and Sellaro, 2000). These issues are likewise critical in the evaluation of the results provided in the present study. This is with a special emphasis toward the systematic lack of information about the course of BD (and BPD). In fact, none of the studies included in our systematic review and meta-analysis provide satisfactory information about a number of otherwise clinically meaningful potential moderators. Namely, the pooled studies did not provide information about atypical or subthreshold mixed features of current bipolar depressive episodes (whenever this was the

16

syndromal case). Neither, they systematically documented the eventual rapid-cycling course of BD in their subsets. Similarly, the pooled samples involving BPD patients failed to provide enough quantitative information about the otherwise clinically suggestive distinctive features of BPD vs. BD (namely, affective instability, rather than mood instability, or different psychometric profiles of impulsivity, or history of childhood trauma or self-injury behavior). In addition, both the BD and BPD subsets documented in the studies included in the present systematic review and meta-analysis almost invariably failed to provide information about current or lifetime treatment response towards pharmacological and non-pharmacological treatments, suicidal behavior, substance abuse, or presence of (hypo-)manic features. Similarly, no sufficient data were provided across the included studies with regard to quantitative measure of predominant affective temperament, which a special emphasis toward cyclothymia, which has nonetheless proposed to be the underpinning between BD (especially BD-II depression) and the “mood/affective” (actually “emotional”) core of BPD (Perugi et al., 2011). Though, it must be remarked that BPD is a complex construct encompassing varying psychopathological dimensions other than the sole “mood/affective” instability or “impulsivity” shared (yet characterized by differential psychopathological and psychometric profiles) between BPD and BD, to encompass also other clinical variables including, but not limited to, childhood trauma, dissociative experience, and self-injury/para-suicidal behavior (Ghaemi and Barroilhet, 2015). This is of particular relevance considering that the aforementioned variables should represent the translation to BPD of the classical constructs originally developed by Robins and Guze (Robins and Guze, 1970) for psychiatric nosological research (namely, “symptoms”, “genetics”, “course”, “treatment response”) (Ghaemi et al., 2014). Ultimately, these intrinsic limitations of the present study hinder the actual appreciation of additional clinically sound and neurobiological (Chen et al., 2010; Pally, 2002) features

17

associated to comorbid BPD and BD, thus warranting future studies to systematically assess these issues. 4.1 Study limitations As afore mentioned, the main limitations of the present study are essentially related to the DSM “pragmatic approach” towards the diagnosis of BD and BPD and the elusive definition of “mental disorder” as a whole nosological entity by either the DSM-IV or the DSM-5 (Stein et al., 2010), the study design and analysis strategy of the included studies, as documented in the quality assessment scale used. Most studies are observational and based on retrospective assessments. The use of retrospective assessment scales with low sensitivity in discriminating self-report history of actual childhood traumata confabulations or dissociative states may have biased results towards an overestimation of such symptom prevalence. Some of the studies do not include a control group. Small sample size and enrollment of subjects mainly from BD-BPD outpatient units may limit generalizability of these results. Potential confounding factors in these studies include demographic and historical illness variables, which often were not appropriately stratified across multivariate modelling. Moreover, in our analyses we anticipated some extent of heterogeneity, which is nonetheless often anticipated when conducting meta-analysis of observational studies (Stoup et al 2000). We therefore attempted to address following the MOOSE guidelines by a) stratifying our results where possible and b) conducting meta-regression analyses. Finally, we also encountered some evidence of publication in our analyses but conducted Trim and fill analyses to diminish the influence of this (Duval and Tweedie, 2000). We acknowledge that the exclusion of original studies providing side-by-side comparison of BD against BPD cases (beyond the mere information about BD↔BPD comorbidity) might have hindered a direct comparison of twos in terms of similarities and differences beyond the sole comorbid cases. Nonetheless, we purposely followed this strategy aiming at enhancing the quality of pooled data. In fact, most of the studies comparing BD↔BPD are qualitative rather

18

than quantitative reports and/or based on chart-review or post-hoc records, meaning that the subsets of the sample(s) at study were almost invariably measured based on fully noncomparable ratings. Specifically, with few notable recent, rigorous exceptions testing a broad ranges of alternative definitions and moderators in non-comorbid cases of either BD or BPD (Bayes et al., 2015), BD cases are simply not systematically assessed for history of childhood trauma or self-injury behavior, otherwise relevant to BPD. Ultimately, this would have led to a major measurement bias invariably affecting the overall validity of quantitative comparisons between BD and BPD as critically discussed over in the text. Nevertheless, allowing for these caveats our study is a first and contains numerous strengths. First, the strength of the selected studies is that the diagnosis of BD and BPD were consistently based on the DSM criteria and were established by trained investigators using validated assessment scales mainly with interrater reliability. The main strength of this review is its being systematic and its including the entire scientific evidence published so far on the main medical databases.

5. Conclusion In our meta-analysis, we established that i) BD and BPD and ii) BPD and BD comorbidity affect approximately one in five people. Further higher rates of BDP-BD comorbidity would nonetheless be expected in the clinical practice beyond the DSM-IV (and DSM-5) approach, ultimately influencing the therapeutic choices and outcomes as well as the accuracy and homogeneity of the diagnostic samples at study for endophenotype and other genetic investigations. Above all, BD-BPD comorbidity is common and necessitates appropriate diagnosis and treatment in clinical practice. Future longitudinal prospective studies are required to better understand the diagnostic boundaries between these two conditions. Comprehensive quantitative assessment of the overlapping and differential clinical moderators is also warranted in order to better understand the actual boundaries of BPD and BD toward the delivery of more

19

accurate therapeutic interventions and a better insight about the potential biomarkers and genetics validators allowing an accurate distinction between the twos.

20

References Agius, M., Lee, J., Gardner, J., Wotherspoon, D., 2012. Bipolar II Disorder and Borderline Personality Disorder co-morbidity or spectrum? Psychiatria Danubina 24 Suppl 1, S197-201. Akiskal, H.S., 1994. The temperamental borders of affective disorders. Acta psychiatrica Scandinavica. Supplementum 379, 32-37. Akiskal, H.S., 2004. Demystifying borderline personality: critique of the concept and unorthodox reflections on its natural kinship with the bipolar spectrum. Acta psychiatrica Scandinavica 110, 401-407. Allen, D.M., Parry, P.I., Purssey, R., Spielmans, G.I., Jureidini, J., Rosenlicht, N.Z., Healy, D., Feinberg, I., 2012. BRIDGE study warrants critique. Archives of general psychiatry 69, 643-644; author reply 644-645. Alnaes, R., Torgersen, S., 1988. The relationship between DSM-III symptom disorders (Axis I) and personality disorders (Axis II) in an outpatient population. Acta psychiatrica Scandinavica 78, 485-492. Amerio, A., Odone, A., Tonna, M., Stubbs, B., Ghaemi, S.N., 2015. Bipolar disorder and its comorbidities between Feinstein and the Diagnostic and Statistical Manual of Mental Disorders. The Australian and New Zealand journal of psychiatry 49, 1073. Angst, J., Sellaro, R., 2000. Historical perspectives and natural history of bipolar disorder. Biological psychiatry 48, 445-457. APA, 2013. Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5). American Psychiatric Publishing, Washington. Barbato, N., Hafner, R.J., 1998. Comorbidity of bipolar and personality disorder. Aust N Z J Psychiatry 32, 276280. Barroilhet, S., Vohringer, P.A., Ghaemi, S.N., 2013. Borderline versus bipolar: differences matter. Acta psychiatrica Scandinavica 128, 385-386. Bayes, A.J., McClure, G., Fletcher, K., Ruiz, Y., Hadzi-Pavlovic, D., Stevenson, J.L., Manicavasagar, V.L., Parker, G.B., 2015. Differentiating the bipolar disorders from borderline personality disorder. Acta psychiatrica Scandinavica. Begg, C.B., Mazumdar, M., 1994. Operating characteristics of a rank correlation test for publication bias. Biometrics 50, 1088-1101. Bridler, R., Häberle, A., Müller, S.T., Cattapan, K., Grohmann, R., Toto, S., Kasper, S., Greil, W., 2015. Psychopharmacological treatment of 2195 in-patients with borderline personality disorder: A comparison with other psychiatric disorders. European Neuropsychopharmacology 25, 763-772. Briere, J., Elliott, D.M., 2003. Prevalence and psychological sequelae of self-reported childhood physical and sexual abuse in a general population sample of men and women. Child abuse & neglect 27, 1205-1222. Chen, G., Henter, I., Manji, H., 2010. Translational research in bipolar disorder: emerging insights from genetically based models. Molecular psychiatry 15, 883-895. Comtois, K.A., Cowley, D.S., Dunner, D.L., Roy-Byrne, P.P., 1999. Relationship between borderline personality disorder and Axis I diagnosis in severity of depression and anxiety. The Journal of clinical psychiatry 60, 752-758. Conus, P., Cotton, S., Schimmelmann, B.G., Berk, M., Daglas, R., McGorry, P.D., Lambert, M., 2010. Pretreatment and outcome correlates of past sexual and physical trauma in 118 bipolar I disorder patients with a first episode of psychotic mania. Bipolar disorders 12, 244-252. DerSimonian, R., Laird, N., 1986. Meta-analysis in clinical trials. Controlled clinical trials 7, 177-188. Downs, S.H., Black, N., 1998. The feasibility of creating a checklist for the assessment of the methodological quality both of randomised and non-randomised studies of health care interventions. Journal of epidemiology and community health 52, 377-384. Duval, S., Tweedie, R., 2000. Trim and fill: A simple funnel-plot-based method of testing and adjusting for publication bias in meta-analysis. Biometrics 56, 455-463. Egger, M., Davey Smith, G., Schneider, M., Minder, C., 1997. Bias in meta-analysis detected by a simple, graphical test. BMJ (Clinical research ed.) 315, 629-634. Feinstein, A.R., 1970. The Pre-Therapeutic Classification of Co-Morbidity in Chronic Disease. Journal of chronic diseases 23, 455-468. Fornaro, M., Aguglia, E., Dell'Osso, L., Perugi, G., 2011. Could the underestimation of bipolarity obstruct the search for novel antidepressant drugs? Expert opinion on pharmacotherapy 12, 2817-2831. Fossati, A., Madeddu, F., Maffei, C., 1999. Borderline Personality Disorder and childhood sexual abuse: a metaanalytic study. Journal of personality disorders 13, 268-280. Fountoulakis, K.N., 2014. Bipolar Disorder: An Evidence-Based Guide to Manic Depression. Springer. Friborg, O., Martinsen, E.W., Martinussen, M., Kaiser, S., Overgard, K.T., Rosenvinge, J.H., 2014. Comorbidity of personality disorders in mood disorders: a meta-analytic review of 122 studies from 1988 to 2010. Journal of affective disorders 152-154, 1-11. Ghaemi, S.N., Barroilhet, S., 2015. Confusing borderline personality with severe bipolar illness. Acta psychiatrica Scandinavica 132, 281-282. Ghaemi, S.N., Dalley, S., Catania, C., Barroilhet, S., 2014. Bipolar or borderline: a clinical overview. Acta psychiatrica Scandinavica 130, 99-108. Goodwin, F.K., Jamison, K.R., 2007. Manic-depressive illness: bipolar disorders and recurrent depression. Oxford University Press. 21

Gunderson, J.G., Stout, R.L., Shea, M.T., Grilo, C.M., Markowitz, J.C., Morey, L.C., Sanislow, C.A., Yen, S., Zanarini, M.C., Keuroghlian, A.S., 2014. Interactions of borderline personality disorder and mood disorders over ten years. Journal of Clinical Psychiatry 75, 829-834. Henry, C., Mitropoulou, V., New, A.S., Koenigsberg, H.W., Silverman, J., Siever, L.J., 2001. Affective instability and impulsivity in borderline personality and bipolar II disorders: similarities and differences. Journal of psychiatric research 35, 307-312. Henry, C., Phillips, M., Leibenluft, E., M'bailara, K., Houenou, J., Leboyer, M., 2012. Emotional dysfunction as a marker of bipolar disorders. Frontiers in bioscience (Elite edition) 4, 2722. Higgins, J., Green, S., 2011. Cochrane handbook for systematic reviews of interventions, version 5.1.0. The Cochrane Collaboration. Ioannidis, J.P., 2009. Integration of evidence from multiple meta-analyses: a primer on umbrella reviews, treatment networks and multiple treatments meta-analyses. CMAJ : Canadian Medical Association journal = journal de l'Association medicale canadienne 181, 488-493. Joyce, P.R., Light, K.J., Rowe, S.L., Cloninger, C.R., Kennedy, M.A., 2010. Self-mutilation and suicide attempts: relationships to bipolar disorder, borderline personality disorder, temperament and character. The Australian and New Zealand journal of psychiatry 44, 250-257. Khalili, N., 2014. A biopsychosocial approach to bipolar-borderline debate: psychological effect of a biological temperament. Acta psychiatrica Scandinavica 130, 158-159. Liberati, A., Altman, D.G., Tetzlaff, J., Mulrow, C., Gotzsche, P.C., Ioannidis, J.P., Clarke, M., Devereaux, P.J., Kleijnen, J., Moher, D., 2009. The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate healthcare interventions: explanation and elaboration. BMJ (Clinical research ed.) 339, b2700. Mackinnon, D.F., Pies, R., 2006. Affective instability as rapid cycling: theoretical and clinical implications for borderline personality and bipolar spectrum disorders. Bipolar disorders 8, 1-14. Maj, M., 2005. ‘Psychiatric comorbidity’: an artefact of current diagnostic systems? The British Journal of Psychiatry 186, 182-184. Maniglio, R., 2014. Prevalence of sexual abuse among children with conduct disorder: a systematic review. Clinical child and family psychology review 17, 268-282. McDermid, J., Sareen, J., El-Gabalawy, R., Pagura, J., Spiwak, R., Enns, M.W., 2015. Co-morbidity of bipolar disorder and borderline personality disorder: Findings from the National Epidemiologic Survey on Alcohol and Related Conditions. Comprehensive psychiatry 58, 18-28. Nagata, T., Yamada, H., Teo, A.R., Yoshimura, C., Kodama, Y., van Vliet, I., 2013. Using the mood disorder questionnaire and bipolar spectrum diagnostic scale to detect bipolar disorder and borderline personality disorder among eating disorder patients. BMC psychiatry 13, 69. Nock, M.K., Kessler, R.C., 2006. Prevalence of and risk factors for suicide attempts versus suicide gestures: analysis of the National Comorbidity Survey. Journal of abnormal psychology 115, 616-623. Pally, R., 2002. The neurobiology of borderline personality disorder: the synergy of "nature and nurture". Journal of psychiatric practice 8, 133-142. Paris, J., Gunderson, J., Weinberg, I., 2007. The interface between borderline personality disorder and bipolar spectrum disorders. Comprehensive psychiatry 48, 145-154. Perugi, G., Akiskal, H.S., 2002a. The soft bipolar spectrum redefined: focus on the cyclothymic, anxious-sensitive, impulse-dyscontrol, and binge-eating connection in bipolar II and related conditions. Psychiatric Clinics of North America 25, 713-737. Perugi, G., Akiskal, H.S., 2002b. The soft bipolar spectrum redefined: focus on the cyclothymic, anxious-sensitive, impulse-dyscontrol, and binge-eating connection in bipolar II and related conditions. The Psychiatric clinics of North America 25, 713-737. Perugi, G., Angst, J., Azorin, J.M., Bowden, C., Vieta, E., Young, A.H., Group, B.S., 2013. The bipolar-borderline personality disorders connection in major depressive patients. Acta psychiatrica Scandinavica 128, 376-383. Perugi, G., Angst, J., Azorin, J.M., Bowden, C.L., Caciagli, A., Mosolov, S., Vieta, E., Young, A.H., Group, B.R.I.-M.S., 2015. Relationships between mixed features and borderline personality disorder in 2811 patients with major depressive episode. Acta psychiatrica Scandinavica. Perugi, G., Fornaro, M., Akiskal, H.S., 2011. Are atypical depression, borderline personality disorder and bipolar II disorder overlapping manifestations of a common cyclothymic diathesis? World psychiatry : official journal of the World Psychiatric Association 10, 45-51. Perugi, G., Toni, C., Travierso, M.C., Akiskal, H.S., 2003. The role of cyclothymia in atypical depression: toward a data-based reconceptualization of the borderline-bipolar II connection. Journal of affective disorders 73, 87-98. Qi, X., Yang, M., Ren, W., Jia, J., Wang, J., Han, G., Fan, D., 2013. Find duplicates among the PubMed, EMBASE, and Cochrane Library Databases in systematic review. PloS one 8, e71838. Reich, D.B., Zanarini, M.C., Fitzmaurice, G., 2012. Affective lability in bipolar disorder and borderline personality disorder. Comprehensive psychiatry 53, 230-237. Robins, E., Guze, S.B., 1970. Establishment of diagnostic validity in psychiatric illness: its application to schizophrenia. The American journal of psychiatry 126, 983-987. Sajatovic, M., Blow, F.C., Ignacio, R.V., 2006. Psychiatric comorbidity in older adults with bipolar disorder. International journal of geriatric psychiatry 21, 582-587. 22

Siever, L.J., Torgersen, S., Gunderson, J.G., Livesley, W.J., Kendler, K.S., 2002. The borderline diagnosis III: identifying endophenotypes for genetic studies. Biological Psychiatry 51, 964-968. Stein, D.J., Phillips, K.A., Bolton, D., Fulford, K., Sadler, J.Z., Kendler, K.S., 2010. What is a mental/psychiatric disorder? From DSM-IV to DSM-V. Psychological medicine 40, 1759-1765. Stroup, D.F., Berlin, J.A., Morton, S.C., Olkin, I., Williamson, G.D., Rennie, D., Moher, D., Becker, B.J., Sipe, T.A., Thacker, S.B., 2000. Meta-analysis of observational studies in epidemiology: a proposal for reporting. Metaanalysis Of Observational Studies in Epidemiology (MOOSE) group. Jama 283, 2008-2012. Swann, A.C., Lijffijt, M., Lane, S.D., Steinberg, J.L., Moeller, F.G., 2013. Antisocial personality disorder and borderline symptoms are differentially related to impulsivity and course of illness in bipolar disorder. Journal of affective disorders 148, 384-390. Vieta, E., Suppes, T., 2008. Bipolar II disorder: arguments for and against a distinct diagnostic entity. Bipolar disorders 10, 163-178. Young, A.H., Angst, J., Azorin, J.M., Vieta, E., Perugi, G., Gamma, A., Bowden, C.L., 2012. BRIDGE Study Warrants Critique-Reply. Archives of general psychiatry 69, 644-645. Zimmerman, M., Mattia, J.I., 1999. Axis I diagnostic comorbidity and borderline personality disorder. Comprehensive psychiatry 40, 245-252. Zimmerman, M., Morgan, T.A., 2013. Problematic boundaries in the diagnosis of bipolar disorder: the interface with borderline personality disorder. Current psychiatry reports 15, 422.

Study Highlights: 

    

Borderline Personaliy Disoder (BPD) is common in bipolar disorder (BD), especially in Type-BD (BD-II). The opposite trend is also true with respect to BD rates in BPD samples. Overall, up to 21.6% of BD samples have comorbid BPD. Higher comorbid BPD in BD were noted in BD II participants (37.7%) and North American studies (26.2%). A higher percentage of males and higher mean age significantly (p<0.05) predicted a lower prevalence of comorbid BPD in BD participants. The adjusted prevalence of BD among people with BPD was 18.5%. BPD-BD comorbidity is common, with approximately one in five people experiencing a comorbid diagnosis. Differences rather similarities between BD and BPD should nonetheless receive greater attention by future replication studies in order to provide further clinically informative insights on the matter.

23

Identification

Figure 1: Adapted PRISMA 2009 Flow Diagram.

Records identified through database searching (N=1424, of which 718 though PUBMED; Scopus=277; Embase=276; PsycINFO=147; Cochrane library=6)

Additional records identified through other sources (n=8 edited books or manual search)

Potential relevant citations (n=85)

Eligibility

Screening

Records after duplicates removed (n=453)

Included

Full-text articles assessed for eligibility (n=52)

Thirty-three records excluded after screening of abstract and title

Ten articles were excluded, as they did not fulfill the inclusion and exclusion criteria. These included eight articles that analyzed the prevalence of BPD in BD without specify the rates for BPD or BD, and two articles described studies performed in specific populations less likely to be representative of the average BD-BPD patient.

Studies included in qualitative synthesis (n=42)

24

Figure 2: Pooled prevalence of BPD in BD participants across all studies of BPD in BD patients.

Study name

Statistics for each study

Event rate and 95% CI

Event Lower Upper rate limit limit Wilson et al 2007 Perugi et al. 2011 Joyce et al. 2004 Benazzi 2008 Zimmermann and Mattia 1999 Preston et al. 2004 Carpiniello et al 2011 Rossi et al 2001 McDermind et al 2015 Comtois et al 1999 Peselov et al. 1995 Loftus and Jaeger 2006 Garno et al 2005 Perugi et al 2013a Barbato and Hafner 1998 Vieta et al 2000 Benazzi 2000 Pica et al. 1990 Benazzi 2002 Ucok et al 1998 Perugi et al 2015 Brieger P. et al., 2003 Vieta et al 2001 Gasperini et al 1993 Dunayevich et al 2000 George et al. 2003 Carpenter et al 1995 Alnaes and Torgersen 1988

0.500 0.470 0.463 0.457 0.429 0.400 0.316 0.296 0.288 0.235 0.234 0.196 0.170 0.145 0.143 0.125 0.120 0.115 0.115 0.100 0.091 0.067 0.062 0.056 0.051 0.038 0.021 0.016 0.186

0.328 0.365 0.319 0.375 0.262 0.253 0.209 0.201 0.266 0.122 0.135 0.109 0.108 0.130 0.066 0.053 0.055 0.038 0.061 0.053 0.068 0.025 0.031 0.018 0.016 0.010 0.001 0.001 0.144

0.672 0.577 0.615 0.540 0.613 0.567 0.447 0.411 0.311 0.405 0.375 0.327 0.257 0.163 0.283 0.267 0.242 0.303 0.207 0.181 0.120 0.165 0.119 0.159 0.146 0.141 0.259 0.211 0.237

-1.00

-0.50

0.00

0.50

1.00

Prevalence BPD

25

Figure 3: Pooled prevalence of BD in BPD participants across all studies of BPD in BD patients.

Study name

Event rate and 95% CI Event Lower Upper rate limit limit

Delito J. Et al. 2001 0.625 Zimmerman MC. and Mattia JI. 19990.373 Gunderson JG. Et al. 2014 0.278 Prasad RB et al. 1990 0.238 Comtois KA. Et al. 1999 0.211 Gunderson JG. Et al. 2006 0.194 Akiskal HS. Et al. 1985 0.170 Hudziak JJ. Et al. 1996 0.161 Skodol AE. Et al. 1999 0.133 Zanarini MC et al. 1998 0.095 Pope HG. Et al. 1983 0.091 Zanarini MC et al. 2004 0.052 Alnaes R. et al. 1988 0.011 Links PS. And al. 1988 0.006 0.166

0.377 0.260 0.223 0.103 0.109 0.144 0.108 0.098 0.096 0.069 0.030 0.031 0.001 0.000 0.114

0.821 0.502 0.340 0.460 0.368 0.255 0.257 0.254 0.182 0.129 0.247 0.084 0.154 0.083 0.235 -1.00

-0.50

0.00

0.50

1.00

Prevalence BD

26

Table 1. Studies assessing BPD comorbidity in BD samples according to our inclusion/exclusion criteria for systematic review. Reference

Study design

Country

Study population

Population-based studies: BPD comorbidity in BD samples (McDermid Longitudinal Canada Non-institutionalized et al., 2015) population-based civilian population study (multi-diagnostics sample) of 50 United states (1172 BD-I; 428 BD-II). Total BD cases, n=1600.

(Barbato and Hafner, 1998)

Population study

Australia

42 BD-I patients in remission coming from the Noarlunga Community Health Service Hospital-based studies: BPD comorbidity in BD samples (Joyce PR, Pharmacological New Post-hoc analysis of 2004) clinical trial Zealand two studies: 195 currently depressed patients (of whom, 19 BD-II) and 135 bulimic females (of whom, 22=BD-II). Total subset n=41 depressed BD-II cases) (Vieta E, Cohort study Spain 129 consecutive 2001) remitted BD-I outpatients (Benazzi, 2002)

Cross-sectional study

Italy

78 BD-II patients

(Perugi et al., 2013a)

Cross-sectional study

Europe, Asia, North Africa

(Comtois et al., 1999)

Case-control study

USA

5632 hospital-based or community psychiatry-based patients with current major depressive episode (BD=1761, no clear-cut distinction between BD-I and BD-II) 278 outpatients with Axis I Disorders (BD, n=34)

(Preston GA, 2004)

Cohort study

USA

35 BD-I patients who participated in two multi-center studies of lamotrigine, were

Sample size of comorbid BD-BPD cases

Diagnosis assessment

Quality *

360 out of 1172 BD-I (9%) 101 out of 428 BD-II (24%). Total BD cases of comorbid BPD, n=461 (28.8%). 6 (14.29%)

DSM-IV

25/31

DSM-IV; IPDE; BSI; Patients’ questionnaire

11/31

6 out of 19 (31.6%) and 13 out 22 (59%). Total=19 out of 41 (46.3%)

DSM-IV

26/31

8 out of 129 (6.2%)

DSM-III; SCIDI; SCID-II; YMRS; HAMD; SADS-C DSM-IV

26/31

DSM-IV; MINI; GAF; HCL-32

24/31

DSM-III-R; SCID-I; SCIDII; HAM-D; BDI; STAI DSM-IV; SCID-I, SCIDII; WURS; MRS; HAM-D;

24/31

9 out of 78 BD-II (11.5%) 256 out of 1761 BD (14.5%)

8 out of 34 BD cases (23.5%) 14 out of 35 (40%)

25/31

24/31

27

(Vieta E, 2000)

Case-control study

Spain

(G. Perugi, 2015)

Cross-sectional study

Europe, Asia, North Africa

(Wilson ST, 2007)

Case-control study

USA

(Garno JL, 2005)

Cross-sectional study

USA

(Rossi A, 2001)

Prospective study

Italy

(George EL, 2003) (Dunayevich E, 2000)

Psychotherapy clinical trial Prospective study

USA

(Gasperini et al., 1993)

Cohort study

Italy

(Loftus ST, 2006)

Cohort study

USA

(Perugi G, 2011)

Prospective study

Italy

USA

retrospectively evaluated for the incidence of BPD 40 BD-II outpatients recruited from primary care psychiatric setting 2811 hospital-based or community psychiatry-based patients with a current major depressive episode. DSM-IV-define BD cases, n=464. Of whom, BD-I=288; BD-II=176. 173 outpatients admitted to a multisite project on mood disorders and suicidal behaviour (143 MDD; 30 BDII) 100 BD patients (73 BDI-I; 27 BD-II), of whom, 95 outpatients; 5 inpatients.

GAF

5 out of 40 (12.5%)

DSM-III; SCIDII; SADS-C

24/31

BD-I subset with comorbid BPD, n=42 (9 % of the BD total) and BD-II, n=17 (3.7%)

DSM-IV; DSM5; GAF

23/31

15 out of 30 (50%)

DSM-IV; BDI; HAM-D; BIS11; B-DHI

23/31

17 out of 100 (17%)

DSM-IV-TR; SCID-I; SCIDII; HAM-D; CTQ; YMRS; Schedule for Affective Disorders and Schizophrenia DSM-III; SCIDP; SCID-II-PQ; HAM-D

22/31

2 out of 52 (3.8%) 3 out of 59 (5.4%)

DSM-III; PDE; SADS-C DSM-III-R; YMRS; HAMD

22/31

3 out of 54 (5.5%)

DSM-III; HAM-D; MRS; SIDP

21/31

10 out of 51 (19.6%)

DSM-IV-TR; SCID.I; SCIDII; HAM-D; CARS-Mania; MSIF DSM-IV-TR and modified criteria for the

21/31

188 patients consecutively admitted to a psychiatric research unit for a major depressive index episode (71 BD; 117 MDD) 52 remitted BD-I patients 59 manic or mixed inpatients followed during a 12-month course of illness 213 lithium treated outpatients (100 MDD and 113 BD, unspecified ratio of BD-I or BD-II cases, of whom only 54 were assessed for the presence of BPD) 51 remitted BD-I patients, of whom, 4 inpatients; 47 outpatients

21 out of 71 (29.6%)

107 consecutive outpatients with major depressive

12 out of 25 BD-II cases (48%), none

22/31

21/31

21/31

28

episode with atypical features. BD-II, n=25 and BD-I=46. Note: the diagnostic codes for BD encompassed non-DSM-IV-TR defined cases too. Overall, “bipolar spectrum” cases comprised 83 patients. (Zimmerman M, 1999)

Cross-sectional study

USA

(Pica S, 1990)

Case-control study

Australia

(Benazzi, 2008)

Prospective study

Italy

(Carpiniello B, 2011)

Cross-sectional study

Italy

(Brieger et al., 2003)

Case-control study

Germany

(Peselow ED, 1995)

Prospective study

USA

(Ucok A, 1998)

Case-control study

Turkey

(Benazzi, 2000)

Case-control study

Italy

409 outpatients recruited by Hospital Department of Psychiatry BD cases, n=28, of whom, BDI=8, BD-II=15 and 5 additional BD-NOS. 26 recent-onset BD patients (unspecified ratio of subtypes) compared with 35 recent-onset schizophrenic patients 138 consecutive remitted BD-II outpatients from private practice 57 outpatients from university community Mental Health Centre: 29 BDI-I and 28 BD-II 177 patients: 117 MDD and 60 BD (of whom, n=44 BD-I manic or mixed cases) reviewed for any DSM-IV-defined personality disorder at time of hospital admission 66 outpatients who had a lifetime diagnosis of BD and met minimum Research Diagnostic Criteria for hypomania (47 BD who successfully recovered from the hypomanic episode) 90 BD-I outpatients compared with 58 control subjects 113 patients: 63 unipolar MDE (MDD and dysthymic disorders) and 50 BD-II MDE recruited

among the BD-I subset. Overall, the subset of “bipolar spectrum” cases with comorbid BPD included 39 patients (47%) 5 out 8 BD-I (62.5%); 5 out of 15 BD-II (33.3%) and 2 out 5 (40%) BD-NOS 3 out of 26 (11.5%)

diagnosis of BD; SCID-I; SCID-II; HAMD; ADDS; SID; HSCL-90

DSM-IV; SCID-I; SCIDII

21/31

DSM-III; SIDP; SCID-P; RPMIP; BDI; BRMS; SAPS; SANS

19/31

63 out of 138 (45.9%)

DSM-IV; SCID-I; SCIDII; GAF

19/31

18 out of 57 (31.6%)

DSM-IV-TR; BIS; AQ; CGI; GAF; MCMI; MMPI

19/31

4 out of 60 BDs (6.7%)

DSM-III-R; FFM-NEO; BRMAS; CDRS; MWT; MMSE

17/31

11 out of 47 (23.4%)

DSM-III; SCIDI; SCID-II; SADS-C; IMPS; CGI

16/31

9 out of 90 (10%)

DSM-III; SCIDI; SCID-II

13/31

6 out of 50 BDs (12%)

DSM-IV; SCID-I; SCIDII; MADRS; GAF

9/31

29

(Carpenter D, 1995)

Cross-sectional study

USA

(Alnaes and Torgersen, 1988)

Cross-sectional study

Norway

in a private practice 23 BD-I patients

289 consecutive outpatients with Axis I assessed for Axis-II comorbidity. BD total, n=30, of whom n=19 BD-I cases.

None.

None.

DSM-III-R; PDE; BPRS; SAS-SR; GAS DSM-III

9/31

9/31

BD: bipolar disorder; BD-I: bipolar disorder type I; BD-II: bipolar disorder type II; BPD=Borderline Personality Disorder; MDE: major depressive episode; MDD: major depressive disorder; NOS: not otherwise specified; DSM: Diagnostic and Statistical Manual of Mental Disorders; IPDE: International Personality Disorder Examination; BSI: Brief Symptom Inventory; FFM-NEO: Five Factor Inventory-NEO; BR-MAS: Bech-Rafaelsen Mania Scale; CDRS: Cornell Dysthymia Rating Scale; MWT: Multiple Choice Vocabulary Test; MMSE: Mini Mental State Examination; PDE: Personality Disorder Examination; BPRS: Brief Psychiatric Rating Scales; SAS-SR: Social Adjustment Scale-Self Report Version; GAS: Global Assessment Scale; BIS-11: Barratt Impulsiveness Scale; AQ: Aggressiveness Questionnaire; MCMI: Millon Clinical Multiaxial Inventory-III; MMPI-II: Multiphasic Personality Inventory-II; YMRS: Young Mania Rating Scale; CTQ: Childhood Trauma Questionnaire; MRS: Mania Rating Scale; PDE: Personality Disorder Examination; SADS-C: Schedule for Affective Disorders and Schizophrenia-Change Version; CARS Mania: Clinician-Administered Rating Scale for Mania; MSIF: Multidimensional Scale for Independent Functioning; SID: Semi-structured Interview for Depression; ADDS: Atypical Depression Diagnostic Scale; HSCL-90: Hopkins Symptoms Check List; IMPS: Inpatient Multidimensional Psychopathology Scale; SCID-P: Structured Clinical Interview for DSM-III-R; RPMIP: Royal Park Multi-diagnostic Instrument for Psychosis; BDI: Beck Depression Inventory; BRMS: BechRafaelsen Mania Scale; SAPS: Scale for the Assessment of Positive Symptoms; SANS: Scale for the Assessment of Negative Symptoms; WURS: Wender Utah Rating Scale; GAF: Global Assessment of Functioning; B-DHI: Buss-Durkee Hostility Inventory; MINI: Mini International Neuropsychiatric Interview; HCL-32: Hypomania Checklist for self-assessment.

Table 2: Studies assessing BD comorbidity in BPD samples according to our inclusion/exclusion criteria for systematic review. Reference

Study design

Country

Study population

Hospital-based studies: BD comorbidity in BPD samples (Comtois et CrossUSA 278 patients (38 al., 1999) sectional BPD, 108 other study personality disorders, 132 no personality disorders) (Zimmerman CrossUSA 409 outpatients M, 1999) sectional recruited by study Hospital Department of Psychiatry BPD cases, n=59.

(Zanarini et al., 1998)

Crosssectional study

USA

(Skodol et al., 1999)

Crosssectional study

USA

(Akiskal et al., 1985)

Prospective longitudinal follow-up

USA

504 patients (379 BPD, 125 other personality disorders) 571 patients (240 BPD) of whose 60 BD (45 BD-I; 15 BD-II) 100 BPD patients

Sample size of comorbid BD-BPD cases

Diagnosis assessment

Quality *

8 (no clear-cut stratification of BD types)

SCID; DSM-III-R

24/31

5 out of 8 BD-I total (62.5% or 8.47% of BPD sample); 5 out of 15 BD-II total (33.3% or 8.47% of BPD sample) and 12 out 28 BD-NOS (40% or 20.34% of BPD sample) 36 BD-II out of 38 total (9.5% of BPD sample)

DSM-IV; SCID-I; SCID-II

21/31

SCID; DSM-III-R

25/31

32 out of 240 (13.3%): 22 BD-I (9.17%); 10 BD-II (4.17%)

SCID; DSM-IV

26/31

17 out of 100 (17%)

DSM-III

23/31

30

(Deltito et al., 2001)

Crosssectional study Crosssectional study Prospective longitudinal follow-up Prospective longitudinal follow-up

USA

16 BPD patients

10 out of 16 (62.5%)

SCID; DSM-III

21/31

USA, Italy

87 BPD patients

14 out of 87 (16.09%)

DSM-III-R

22/31

USA

39 patients (33 BPD)

3 out of (9.09%)

DSM-III-R

24/31

USA

15 out of 290 (5.17%)

SCID; DSM-III-R

24/31

(Gunderson et al., 2006)

Prospective longitudinal follow-up

USA

38 out of 196 (19.39%): 23 BD-I (11.73%) and 15 BD-II (7.65%)

DSM-IV

24/31

(Gunderson et al., 2014)

Prospective longitudinal follow-up Prospective longitudinal follow-up Crosssectional study

USA

362 patients (290 BPD, 72 other personality disorders) 629 patients (196 BPD, 433 other personality disorders) 223 BPD patients (161 MDD, 34 BD-I, 28 BD-II) 88 BPD patients

62 out of 223 (31.63%): 34 BD-I (15.25%); 28 BD-II (12.6%) None.

SCID; DSM-IV

24/31

DSM-III

10/31

None.

DSM-III

9/31

Crosssectional study

USA

289 consecutive outpatients with Axis I assessed for Axis-II comorbidity. BPD=44. 21 BPD patients

5 out of 21 (23.8%): 4 BD-I (16.7%); 1 BD-II (4.8%)

DSM-III

9/31

(Hudziak et al., 1996) (Pope et al., 1983) (Zanarini et al., 2004)

(Links et al., 1988) (Alnaes and Torgersen, 1988)

(Prasad et al., 1990)

Canada

Norway

BD: bipolar disorder; BD-I: bipolar disorder type I; BD-II: bipolar disorder type II; BPD=Borderline Personality Disorder; MDE: major depressive episode; MDD: major depressive disorder; NOS: not otherwise specified; DSM: Diagnostic and Statistical Manual of Mental Disorders (Third [DSM-III], Third-Revised [III-R], Fourth [DSM-IV], Fourth, Text-Revised [DSM-IV-TR] or Fifth [DSM-5] editions). SCID: Structured Clinical Interview for Mental Disorders.

Table 3: Sub-group analyses results investigating BPD comorbidity in BD. Analysis

Number of study estimate s

Meta-analysis

Prevalenc e BPD

Main analysis

28

18.6%

95% CI

14.4 4

Heterogeneit y

Betwee n group p value

23.7 1

I2

91%

BD type BD I

7

12.46%

6.97

21.2 7

80

BD II

6

27.26%

16.6

41.2

89

Publication bias

Egger Trim and test (p fill (95% value) CI) [adjusted studies] -0.88, 21.6% p=0.3 (17.03 27.1)[5]

0.10 -2.7, p=0.2 3 -5.1,

13.1% (6.7-24.1( [1] 37.7% 31

BD mixed

Geographic al region Europe

15

18.49%

6

8

13.1 2

25.4 2

90

p=0.1 2 -0.95, p=0.5 1

(21.956.6) [2] 20.6% (15.127.5) [2]

-0.39, p=0.0 8 -0.23, p=0.3 5 -7.7, p=0.2 8 N/A

16.8% (9.9-27.2) [1] 26.2% (18.735.3) [2] Unchange d

-5.21, p=0.6 8 -2.1, p=0.2 1 N/A

Unchange d

0.49 12

16.55%

10.9 1

24.2 9

90

North America

11

22.07%

14.5 4

32.0 4

78

Oceania

3

22.84%

10.0 1

44.0 6

89

Various

2

11.56%

4.44

26.8 9

88

Setting Inpatient

4

11.83%

5.22

24.6 5

82

Outpatient

20

20.79%

15.1 6

27.8 3

90

Mixed

2

18.21%

6.65

41.0 5

0

N/A

0.49

22.1% (16.129.5) [2] N/A

Key: N/A = not applicable, BD= bipolar disorder, BPD= borderline personality disorder.

Table 4: Sub-group analyses results investigating BD comorbidity in BPD. Analysis

Number Meta-analysis of study estimate s Prevalenc 95% CI e BD

Main analysis 14

16.58%

11.3 8

Heterogeneit Publication bias y

Betwee n group p value

I2

Egger test (p value)

87

-0.85, p=0.6 1

0

-1.67, p=

77

N/A

17.9% (11.925.7) [1] N/A

91

-0.69, P=0.8

Unchange d

23.5 3

BD type BD I

Trim and fill (95% CI) [adjusted studies] 18.5% (12.726.1) [2]

0.24

3

15.30%

6.47

2

12.65%

8

19.89%

4.79 12.2 3

BD II BD mixed

32.0 6 29.4 7 30.6 7

32

0 Geographic al region Europe

0.19

1 North America+ Europe North America

1

12

1.11%

4.87

20.5 6

16.09%

3.91

47.4 6

17.50%

11.7 0

25.3 6

Setting Inpatient

0

N/A

N/A

0

N/A

N/A

89

-0.37, p=0.8 5

18.8% (12.527.1) [1]

0

N/A

N/A

88

Unchange d

0

-0.53, p=0.8 6 N/A

80

N/A

N/A

0.65 1

9.50%

2.23

19.34%

11.9 6

32.6 1

Outpatient 9 Mixed 2

14.60%

5.36

Unclear 2

10.46%

2.40

29.7 2 34.0 8 35.7 6

N/A

Key: N/A = not applicable, BD= bipolar disorder, BPD= borderline personality disorder, BD= bipolar disorder. Appendix A: MEDLINE search strategy. SET 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

MEDLINE Bipolar disorder BD Bipolar Manic depressive disorder Manic depressive Manic Bipolar disorder Sets 1-7 were combined with “OR” Borderline Personality Disorder Borderline BPD Borderline personality disorder Sets 9-12 were combined with “OR” Sets 8 and 13 were combined with “AND” Set 14 was limited to November, 4 2015, Humans, English language, Adult: 19+ years

Words written in italic were used as MeSH headings, the others were used as free text.

Contributors

33

MF and BS conceived the study and extensively edit the main-text and its attachments. LO, SM, DDB, GP, AV, LG, MS and NV assisted either in manuscript revisions, preparations of the tables and figures or interpretation of results. Finally, all the co-authors substantially contributed to the present piece of work before approving it for final submission.

Role of the funding source There is no funding source to disclose in conjunction with the present piece of work.

Acknowledgments None to state.

34